# Multi-GPU Optimization

CogVideoX is an open-source video generation model originating from Qingying. This version is a dedicated image-to-video model based on a 5B parameter scale.

This model is a summary generation model trained from scratch on an unknown dataset, primarily used for biomedical text processing tasks.

An automatic speech recognition (ASR) model fine-tuned on the Common Voice 16.0 English dataset based on w2v-bert-2.0

A speech recognition model fine-tuned based on OpenAI Whisper-large-v3, specializing in Air Traffic Control (ATCO) scenarios with a word error rate of 17.04%

Dialogue model fine-tuned on the ultrachat_200k dataset based on Mistral-7B

This model is a fine-tuned version of openai/whisper-base on an Arabic Quran dataset, specializing in transcribing Quranic content in Arabic.

BioBERT is a BERT-based pretrained language model specifically optimized for biomedical text mining tasks, suitable for question answering systems.